Social
Impact of Vietnam-era G.I. Bill Eligibility on the Distribution of HbA1c Among Older Americans: Evidence from the Vietnam Draft Lottery Natural Experiment Aayush Khadka* Aayush Khadka Lucia Pacca Abigail Arons Anusha M. Vable
Diabetes complications risk increases non-linearly with increasing glycosylated hemoglobin (HbA1c) levels. Structural interventions that reduce fundamental causes of disease (e.g., low education) may reshape the HbA1c distribution by having larger protective effects at higher HbA1c levels. We assessed this theory in the context of the Vietnam-era G.I. Bill, a structural intervention that subsidized college education for 6 million veterans. Using the Vietnam draft lottery natural experiment, we estimated the causal impact of G.I. Bill eligibility on the HbA1c distribution among Health and Retirement Study respondents. Eligibility depended on length and manner of discharge from military service. Our analytic sample included men born between 1947-1953 with one valid outcome measure (N=1,639). We proxied G.I. Bill eligibility using lottery-defined draft eligibility (intention-to-treat [ITT] analysis) and self-reported veteran status (instrumental variables [IV] analysis; instrument: draft eligibility). HbA1c measured from blood spots was our primary outcome. We estimated ITT effects using linear and quantile regressions and IV effects using two-stage least squares and IV quantile regression estimators. We adjusted for year-and-month-of-birth fixed effects and estimated heteroskedasticity robust standard errors in all models. Draft eligibility lowered mean HbA1c (b=-0.09, 95% confidence intervals (CI) -0.23,0.05), had negligible impact on lower HbA1c quantiles, and had protective effects at higher quantiles (e.g., 10th quantile: b=0.00 [95%CI -0.08,0.08]; 90th quantile: b=-0.25 [95%CI -0.51,0.01]). IV results were generally in the same direction as the ITT effects, but standard errors were larger (e.g., 10th quantile: b=-0.13 [95%CI -1.30,1.03]; 90th quantile: b=-1.39 [95%CI -31.44,28.66]). Results suggest that G.I. Bill eligibility reshaped the HbA1c distribution to one of lower diabetes risk by being more effective in the high-risk right tail of the HbA1c distribution.
